Light rail in America is about as dangerous as the automobile trips they replace and twice as dangerous as the bus trips they replace, according to the data I’ve found on the subject. While I intuitively figured that my favorite mode— elevated rail— was safer, I began this research blissfully believing that transit was always safer than cars. Not true, apparently. The good news is that some LRT systems are much, much safer than others, with design the most important factor.

With so many cities thinking about light rail, you’d think LRT safety would get more attention. But, as with roads, we tend to install light rail designs and then worry about mitigating safety problems later when it’s really too late to change the design. I’m not enthusiastic about the evidence cited below and would enjoy counter-arguments. (Btw, everyone agrees that transit passengers are safe on board LRT; this is about the dangers to drivers, pedestrians, cyclists, and people who are suicidal.)

John Semmens is based at the Arizona Transportation Research Center. I take no pleasure in linking to an article at the Foundation for Economic Education, but here we are. (Similar Semmens here.) Skip his distracting use of vehicle miles of travel and get just pay attention to passenger miles:

The aggregate fatality rate for auto travel is around 15 persons per billion vehicle-miles of travel. However, this includes rural travel, where the fatality rate per billion vehicle-miles is 23. The nationwide fatality rate per billion vehicle-miles of urban automobile travel is 11, and when passengers are included, it drops to 10. Thus we find that light rail’s 14 fatalities per billion passenger-miles of travel and commuter rail’s 12 fatalities per billion passenger-miles of travel are actually higher than the rate for privately operated automobiles.

Also note that Semmens— and everyone else— agrees that buses and grade-separated rail are safer than driving, almost twice as safe. I believe he’s right to use “urban auto” trips since that’s what LRT replaces. And I’ll add that a disproportionate number of auto fatalities occur in the wee hours when most light rail systems sleep.

Todd Litman of the Victoria Transit Policy Institute finds similar numbers for Heavy Rail, Transit Buses, Commuter Rail, and Passenger Cars in this pro-transit article entitled “Safer Than You Think”. (See chart on page 11.) And note that he has no corresponding bar for Light Rail. In the first p. 10 chart, he lumps light in with subways for “passenger fatalities”. Maybe Litman has some reason for doing this, but it would have been a good opportunity to refute Semmens if he could.

Jeff Sabatini with Car and Driver Magazine has good reason to be pro-auto, but his data seems to really have it in for light rail. Cute little chart here about fatalities with various vehicle modes, but compared to 1.1 fatalities per 100 million passenger miles for Passenger Cars, Sabatini gets these numbers for the following modes: Motorcycles (31.5) Light Rail Transit (22.6) Commuter Rail (10.8) Transit Bus (3.7) Heavy Rail Transit (1.4). Here, light rail is defined as “streetcars, tramways, and trolleys”.

An anonymous blogger, writing for a website connected to a shared living home near Carleton College, 40 miles south of Minneapolis, looked at US government data for 1990-97 and concluded that “light rail is indeed less safe than automobiles, but only barely.”

Again, if there’s data to the contrary, I’d enjoy seeing it. But even if light rail, on average, in the United States is about as safe as automobiles, that’s only half the story because there are extremely different fatality records among urban rail systems— and the distinction appears to be less between light and heavy rail, but rather in whether trans are segregated from drivers, pedestrians, and cyclists.

Dug Begely, writing for the Houston Chronicle, made this chart to compare how well Houston Metro was doing on rail safely when compared to other cities.

Houston’s street-running rail is simply inherently unsafe. But you don’t have to take my word for it.

“Each community and each system is different, and there are probably different factors,” said Metro CEO Tom Lambert… He noted Houston’s system, unlike some others, has more at-grade street crossings and doesn’t have the level of restrictions other systems have to keep trespassers off the tracks. That makes safety a larger part of Metro’s day-to-day challenges, he said.

I couldn’t find a well-researched journal article that was newer & more on-topic than this 2001 document by the Federal Transit Administration’s Transportation Research Board. Korve, et al., go into the weeds on fatalities and accidents, comparing 11 light rail systems in the US and Canada. It’s terrific if you want to think deeply about the design of light rail alignments, but there’s not a handy quote to pull out for this article. Clearly, though, Metrolink in St. Louis, with its Exclusive and Semi-Exclusive design did well in its first two years and apparently continues to be largely fatality-free. Ditto Edmonton. But even at the time, Los Angeles’s Blue Line was notable for an unacceptable number of fatalities on surface LRT.

And it’s not the transit agency’s management that seems to matter so much. LA’s Green Line is about the same length as the Blue Line, with 14 stations compared to the Blue LIne’s 22, carries almost half the ridership, and opened in 1995, five years after the Blue Line. The Wikipedia article on the Green Line claims that it’s had only one fatality. I think it’s more like a half-dozen. But that’s a small fraction of the 120+ Blue Line fatalities. The Green Line is entirely above grade, running mostly in the median of Interstate 105. The Blue Line has some underground and elevated sections, but its street-running segments have been deadly.

The most thoughtful comparison of at-grade LRT with non-grade options might be this blog article from “Voony” in Vancouver. Drawing mostly from French data comparing tramways with subway, he concluded that subways are an order of magnitude safer than LRTs. US data seems to be pretty similar.

Non-grade catenary light rail is apparently safer than third rail-powered heavy rail. Accidental deaths on Chicago Transit Authority elevated lines sometimes were the result of a passenger or employee slipping down onto the rails or even crossing the rails. They weren’t hit by trains, but instead came in contact with the third rail and were electrocuted. Platform screens reduce suicides and such accidents to nearly zero. Tokyo plans to gradually install platform screens, but is also trying blue lighting on platforms to reduce suicides. 

All in all, at this point, I agree with “Voony” that surface LRT has about ten times the number of accidents as non-grade rail— and about three times the number of fatalities. Anyone advocating replacing buses with street-running LRT should be called upon to defend a mode that’s about twice as dangerous. While trains can sometimes be acceptably segregated at grade, Vision Zero for rail is grade separation with platform screens.

Update Aug 29: Also keep an eye on this forthcoming Transportation Research Board study from Eric Dumbaugh and Candace Brakewood:

Conventional wisdom holds that transit is safer than other forms of surface transportation. But light rail produces 14.8 fatalities for every billion passenger miles, making it 1.5 times more dangerous than personal automobiles, and more than 17 times more hazardous than bus service. It should further be observed that these numbers only account for fatal crashes involving light rail vehicles…

This study will entail a two-tiered analysis. The first analysis will entail a systematic examination of national trends in light rail safety performance. The second tier will investigate the design-level factors that may be responsible to the crash risk associated with light rail transit.

This study will provide a detailed analysis of the safety impacts of two recently-developed light rail systems in the U.S.: Orlando SunRail and Charlotte Lynx.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes:

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>